# MCP Atlassian — Jira + Confluence Tools for Agents

> mcp-atlassian is an MCP server for Jira and Confluence so your agent can search, create, and update issues/pages via IDE MCP configuration.

## Install

Merge the JSON below into your `.mcp.json`:

## Quick Use

1. Create an Atlassian API token (README links to the API token page).
2. Add MCP config (Claude Desktop / Cursor example from README):
   ```json
   {
     "mcpServers": {
       "mcp-atlassian": {
         "command": "uvx",
         "args": ["mcp-atlassian"],
         "env": {
           "JIRA_URL": "https://your-company.atlassian.net",
           "JIRA_USERNAME": "your.email@company.com",
           "JIRA_API_TOKEN": "your_api_token",
           "CONFLUENCE_URL": "https://your-company.atlassian.net/wiki",
           "CONFLUENCE_USERNAME": "your.email@company.com",
           "CONFLUENCE_API_TOKEN": "your_api_token"
         }
       }
     }
   }
   ```
3. Ask your assistant to search issues or pages (README includes examples).

## Intro

mcp-atlassian is an MCP server for Jira and Confluence so your agent can search, create, and update issues/pages via IDE MCP configuration.

- **Best for:** teams who live in Jira + Confluence and want agents to read/write with auditability through tool calls
- **Works with:** Claude Desktop or Cursor MCP configs, uvx runner, Atlassian Cloud API tokens or Server/Data Center personal tokens
- **Setup time:** 15–30 minutes

## Practical Notes

- README claims 72 tools total and includes a compatibility table (Cloud + Server/Data Center versions).
- Config example uses uvx and environment variables for Jira and Confluence credentials.

## How to Keep It Safe in Production

Jira/Confluence write access is powerful. Guardrails:

- Use least-privilege credentials (separate service account if possible).
- Keep tokens out of the repo; load them via your MCP client’s env management.
- Start with read-only workflows (search, get issue/page) before enabling create/update flows.

## Common “Good” Agent Tasks

- Find issues assigned to a user via JQL, then summarize into a standup note.
- Search Confluence for onboarding pages and compile a checklist for new hires.


### FAQ

**Q: Does it support on-prem Atlassian?**
A: Yes—README states Server/Data Center is supported and notes using personal access tokens.

**Q: How is it configured?**
A: Via MCP client config; README shows uvx command and env var setup.

**Q: How many tools are included?**
A: README states 72 tools total (see Tools Reference in docs).

## Source & Thanks

> Source: https://github.com/sooperset/mcp-atlassian
> License: MIT
> GitHub stars: 5,153 · forks: 1,160

---

<!-- ZH -->

## 快速使用

1. 创建 Atlassian API token（README 提供 token 页面链接）。
2. 添加 MCP 配置（README 的 Claude Desktop / Cursor 示例）：
   ```json
   {
     "mcpServers": {
       "mcp-atlassian": {
         "command": "uvx",
         "args": ["mcp-atlassian"],
         "env": {
           "JIRA_URL": "https://your-company.atlassian.net",
           "JIRA_USERNAME": "your.email@company.com",
           "JIRA_API_TOKEN": "your_api_token",
           "CONFLUENCE_URL": "https://your-company.atlassian.net/wiki",
           "CONFLUENCE_USERNAME": "your.email@company.com",
           "CONFLUENCE_API_TOKEN": "your_api_token"
         }
       }
     }
   }
   ```
3. 按 README 的示例，让助手开始搜索/创建/更新工单与页面。

## 简介

mcp-atlassian 是面向 Jira/Confluence 的 MCP server（支持 Cloud 与 Server/Data Center）：通过 IDE 的 MCP 配置让 Agent 能搜索、创建与更新工单/页面，把知识库与工单系统接入工具层。

- **适合谁：** 深度使用 Jira + Confluence、希望 Agent 通过工具调用实现可审计读写的团队
- **可搭配：** Claude Desktop / Cursor 的 MCP 配置、uvx 运行方式、Atlassian Cloud API token 或 Server/Data Center 的个人令牌
- **准备时间：** 15–30 分钟

## 实战建议

- README 标注总计 72 个工具，并提供 Cloud 与 Server/Data Center 的兼容性表格。
- 配置示例使用 uvx，并通过环境变量注入 Jira/Confluence 凭据。

## 上生产前的安全护栏

Jira/Confluence 的写权限很强，建议：

- 最小权限账号：最好独立服务账号，权限只给必须项。
- Token 不进仓库；通过 MCP 客户端的 env 管理注入。
- 先跑只读流程（search/get），再逐步开放 create/update。

## 典型任务范式

- 用 JQL 拉取某人负责的 issues，并汇总成 standup note。
- 在 Confluence 里搜索 onboarding 文档，整理新同事 checklist。


### FAQ

**支持自建部署吗？**
答：支持。README 说明兼容 Server/Data Center，并提示使用个人访问令牌。

**怎么配置？**
答：通过 MCP 客户端配置；README 给出 uvx 启动与 env 变量示例。

**工具数量多少？**
答：README 标注总计 72 tools（详见文档 Tools Reference）。

## 来源与感谢

> Source: https://github.com/sooperset/mcp-atlassian
> License: MIT
> GitHub stars: 5,153 · forks: 1,160


---
Source: https://tokrepo.com/en/workflows/mcp-atlassian-jira-confluence-tools-for-agents
Author: MCP Hub